detecting "shit" frames (Q-update :)

[lwk.ms]

Hi, to the crowd,

this is my first posting on doom9.net, a site which helped/helps me A LOT! I'm afraid i must bother you with a B I G THANK YOU! All of you, especially doom9 and all aides which made doom9.net a reference for some special encoding needs!

This had to be said!

Now for my Q (short-version at the end):

This Q was asked before (2001-09-13) and i wondered if there is a new answer to the problem.
The problem ? How can can i be sure that the "anti-shit"-option (i use "latest" nandub 1.0rc2) really worked, besides from watching the whole (ripped) movie again ?
Given, that the source, the vobs, are still available, wouldn't it be possible for a tool, to compare individual frames (*.avi + *.vob) using a similar technique as the "anti-shit"-option in nandub does ?
maybe a bit faster ? ;)
i think, i would appreciate e.g. 30 more minutes for such a tool to check if everything is allright (really NO shit) with the movie ?!

Some of you might say, that if the nandub set-up is correct, there won't be shit in the movie, but maybe it's save to say... shit happens ?

anyway, i just used antishit with the values 16db, 0db, 25% and i think these settings are ok for eliminating shit, although i'm not quite sure if i really need the 25% after setting the min quality to "0" (which means the keyframe is a real keyframe and not a compressed delta frame ?!)

ok, in short : Is there, in the meantime, any tool/solution available to make sure (fast) that there is no "shit" in the movie ?

Answers and/or suggestions are very welcome!

Maybe it's just me, not seeing the "is-there-any-"shit"-in-this-movie-output-log-button" which is hiding somewhere ...

Thx !

Martin

[LotionBoy]

There is no tool to do that.

from my experience with anti-shit settings of 16,1,0, there is no shit. Which doesn't mean that it is always true, but does mean that on a rather large sample size, it is true. Have you had problems with anti-shit not fixing problems?

LotionBoy

[lwk.ms]

> There is no tool to do that.

So, i suppose nothing changed 'til then... shit...

> 16, 1, 0

correct me if i'm wrong:

the "anti-shit"-option determines which "similarity-level" the encoded frame has, compared to the source-frame on a scale from 1db to 95db (0 disables "anti-shit") whereby "1" would represent the most differences and "95" the least. So if the encoded frame has less than 16db it is reencoded as a normal keyframe.

the "min. quality" appears to do nearly the same, except that it doesn't create real keyframes, but less compressed delta-frames.
it is not clear to me, why i should use this... i don't think that "shit" happens very often and therefore, if a frame would contain "shit" i replace the broken frame with a keyframe (with "anti-shit") and the resulting delta-frames wouldn't contain shit anymore ?!
with your "1" setting, it seems you'd like to have it disabled as well ?!

the "motion modulation"-option tries to compensate the frame-differencies in high-motion-scenes. Such scenes are naturally more different from the source-frame because of more macro-blocks ?!
So i think, that this option controls "anti-shit" to not create too much keyframes in high-motion-scenes.

so, am i wrong? and if i'm not, why did you choose 16,1,0 ?

> Have you had problems with anti-shit not fixing problems?

not so far, because i'm starting to use it now... :)
in earlier rips, i only set "anti-freeze".
but i am now on ripping blade-runner (german rc2) and realized that shit is really annoying !
so i ripped it once with anti-freeze to a size of 542mb, messed a little bit with the *.vobs (using ifo-edit, first-time) deleted the vobs and saw the shit at frame 86822 (deckard has just "retired" zhora and is buying something to drink).so i rerent the dvd and wanted to do it right this time.
i watch the new encoding (16,0,25) now, to manually detect if there's any "shit" ;)

Martin
20:00, 8pm

[lwk.ms]

hi,

so i watched it now, real carefully and the annoying "shit" is gone but it's still some left...
in one scene, an overview over a marketplace with camera-movement downwards to deckard, handing out the snake-scale for electron-microscope-research, there is some shit left, mainly in dark/black areas and hardly noticeable...
i encoded this part again starting 56 seconds earlier using the 16,30,50 setting and now it's gone. but, i think this is no guarantee because this clip (duration 2min 7sec) has not the same keyframe-arrangement.
so... i'm going for a third time :D.

that's what i meant...